A Neurobiological Investigation of Visual Target Detection and the Optic Lobe of Dragonflies

Abstract

For many decades the insect nervous system has provided novel insights into the mechanisms of visual processing. Despite commonly being labelled as ‘simple’, recent evidence suggests that some insects have remarkably complex brains. Guided by a brain the size of a poppy seed, dragonflies detect and pursue prey amongst cluttered backgrounds with high success rates (Olberg et al. 2000; Combes et al. 2013). These pursuits are not simple reactionary processes, instead the brain uses internal models and selective attention to maximise performance in challenging conditions (Wiederman and O’Carroll 2013; Mischiati et al. 2015). However until now we have no detailed and up-to-date description of how the dragonfly optic lobe is organised, and little understanding of the strategies optic lobe neurons use to detect and track visual features. My initial work describes the morphology and organisation of the dragonfly optic lobe, the most complex optic lobe of any insect studied to date. I demonstrate that in contrast to recent reports, the dragonfly lobula complex differs substantially from its dipteran counterparts. Furthermore, both the second and third optic ganglia contain approximately twice as many synaptic layers as any other insect. Next I performed a series of electrophysiological experiments that investigated the effects of target trajectory on the responses of target-detecting neurons. A small feature drifting across the retina generates a weak and variable signal. For this reason the human brain has adopted a strategy where target movement is integrated across a trajectory in a predictive manner, improving signal strength while ignoring distractors (Watamaniuk et al. 1995). I demonstrate that a facilitation mechanism modulates gain across the receptive field of target-detecting neurons, maximising responses to targets presented at a predicted location and suppressing responses to targets elsewhere. This modulation of gain results in large improvements in local contrast sensitivity, and also induces strong direction selectivity that matches the direction of stimuli in the recent past. I then investigated how different parameters of a targets trajectory affect the intensity and spatial spread of gain modulation. Targets of differing velocity, contrast, size, duration and trajectory length were drifted through the receptive field, before quantifying the strength of gain modulation. I show that gain modulation is gated by target contrast, and that the magnitude of modulation is dependent on complex interactions between the parameters of a primers trajectory and the probe that follows. Finally, I investigated whether this gain modulation was a mechanism underlying the selective attention previously reported in target-detecting neurons (Wiederman and O’Carroll 2013). When presented with two targets simultaneously, target-detecting neurons select and respond to one, and ignore the presence of the other. With the use of a novel frequency-tagging stimulus, I demonstrated that when presented with two competing targets, both the selected and unselected target trajectories induced an increase in gain ahead of their path. This result suggests that predictive gain modulation is not a mechanism of selective attention, but a parallel processing strategy that acts to improve signal strength in challenging conditions. Finally, I characterised the physiological responses of a population of neurons sensitive to the movement of larger features or patterns. Controlling flight at highspeed would benefit from the detection of low frequency patterns in the environment. We show that bar-sensitive neurons in the dragonfly lobula complex have highly abnormal responses to stimulus velocity. Our data describes the diverse physiological properties of these neurons, including their tuning for motion direction, height, width and velocity, and their sensitivity to contrast of different polarities. Together, my thesis provides a significant contribution to our knowledge of the visual system of dragonflies. In a broader sense, these findings build our understanding of the structure and function of nervous systems, and the strategies implemented to efficiently solve challenging sensory tasks such as small target detection.